Energy Strategy Reviews 31 (2020) 100526
Available online 28 July 2020
2211-467X/© 2020 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Modeling natural gas consumption, capital formation, globalization, CO 2 emissions and economic growth nexus in Malaysia: Fresh evidence from combined cointegration and causality analysis
Mfonobong Udom Etokakpan
a,b, Sakiru Adebola Solarin
c, Vedat Yorucu
a, Festus Victor Bekun
d,e, Samuel Asumadu Sarkodie
f,*aDepartment of Economics, Famagusta, Eastern Mediterranean University, North Cyprus, Via Mersin 10, Turkey
bEconomics Department, Babcock University, Ogun State, Nigeria
cCentre for Globalisation and Sustainability Research, Multimedia University, 75450, Melaka, Malaysia
dFaculty of Economics Administrative and Social Sciences, Istanbul Gelisim University, Istanbul, Turkey
eDepartment of Accounting, Analysis and Audit School of Economics and Management South Ural State University, 76, Lenin Aven., Chelyabinsk, 454080, Russia
fNord University Business School (HHN). Post Box 1490, 8049, Bodø, Norway
A R T I C L E I N F O Keywords:
Natural gas consumption Globalization index Economic output Combined cointegration Malaysia
A B S T R A C T
The discovery of natural gas in the 20th century has increased aggregate energy consumption while spurring economic development. However, very little attention has been given in the energy economics literature, especially in Malaysia. As such, this paper primarily revisited the natural gas — economic growth nexus hy- pothesis in the case of Malaysia. The study was conducted with data from 1980 to 2014 in a multivariate framework with the inclusion of capital formation, globalization, and CO2 emissions to avoid omitted variable bias. We investigated the stationarity properties with a method that accommodates a single structural break.
Subsequently, the novel combined co-integration test in conjunction with several techniques were used to assess the magnitude of the long-run equilibrium relationship. The empirical findings trace the long-run equilibrium relationship among the variables over the sampled period. The Granger causality test analysis confirmed the growth-energy driven hypothesis in Malaysia. The findings call for the adoption of cleaner and environmentally friendly energy sources in the Malaysian energy mix. We highlight the need for pragmatic strides from both private and public energy sector stakeholders to prioritize clean and accessible energy in line with the Sus- tainable Development Goals.
1. Introduction
Energy is identified as an integral driver of socio-economic devel- opment of all forms of economies — developing, transition, and devel- oped economies [1]. The last two decades have experienced a persistent demand for energy sources like natural gas, oil, electricity consumption across the globe [2]. The continuous and persistent pressure for more energy sources puts pressure on the environment. This has been a heated debate among environmental economist, stakeholder and policymakers that design and formulate energy strategies [3–7].
Energy sources could either be from fossil fuel sources like crude oil, coal, and uranium or renewable energy sources like solar, geothermal, biomass, hydro, and wind, which are the alternative cleaner energy
sources. However, non-renewable energy sources are known to emit carbon dioxide emissions which translate into environmental deterio- ration. The environmental and health-related hazards attributed to fossil fuels have raised concern and discourse among nations. Thus, in energy- dependent economies, there is a potential tradeoff between productivity and environmental sustainability [5]. Natural gas (NG hereafter) is somewhat preferred among other fossil fuel energy sources due to its low carbon intensity and limited environmental effects in production and consumption compared to oil and coal [8]. Global energy demand for natural gas rose from 5353 Tcf in 1980 to 113 trillion cubic feet (Tcf) in 2010. This swift increase was experienced across regions in the world.
For instance, in the Middle East, NG demand rose from 3.1 Tcf in the 1980s to an overwhelming 51.7 Tcf in 2017. This sharp increase is
* Corresponding author.
E-mail addresses: [email protected] (M.U. Etokakpan), [email protected] (S.A. Solarin), [email protected] (V. Yorucu), festus.
[email protected] (F.V. Bekun), [email protected] (S.A. Sarkodie).
Contents lists available at ScienceDirect
Energy Strategy Reviews
journal homepage: http://www.elsevier.com/locate/esr
https://doi.org/10.1016/j.esr.2020.100526
Received 22 August 2019; Received in revised form 28 June 2020; Accepted 16 July 2020
attributed to the rapid economic expansion initiated in the region. A similar trend of increased consumption of NG in North America was observed from 58.5 Tcf in 1980 to 91.2 Tcf in 2017 [9]. Thus, natural gas production and consumption play a pivotal role in economic prosperity across countries [10,11].
Several studies have since emerged after the seminal work on energy- growth nexus [12] documented in the existing literature (see: [13–16].
The studies outlined incorporate other interesting variables, however, there is no consensus in the reported direction of causality. The plausible explanation for the divergent empirical findings could be due to varying sample size selection, estimation techniques (methodology), and selected sample area. Regarding natural gas consumption, there exists a paucity of studies (see [10,11,17]: for the Asian countries, however, Malaysia has received little attention. Malaysia is worthy of investiga- tion, given its energy portfolio with almost 40% share of energy con- sumption from natural gas (see Fig. 1).
To this end, we assess the nexus between NG consumption and economic growth with a novel perspective in a multivariate framework.
This is necessary to avoid the omitted variable bias which earlier studies failed to address. To circumvent the issues of omitted variable bias, we incorporate capital formation, globalization, and CO2 emissions as additional variables. For the case of Malaysia, few studies [18,19] exist in the literature. Thus, we seek to improve the literature on natural gas energy and limited studies in Malaysia by investigating the theme via new insights that account for useful variables. This is crucial, given the key role of natural gas consumption in the Malaysian economic output.
The study further strengthens the quest to achieve the 11th Malaysian Plan and goal 7 of the sustainable development target. In terms of estimation method, a novel combined non-cointegration, Granger cau- sality, and Zivot and Andrews unit root techniques are utilized to examine the cointegration, causal direction, and account for a single structural break.
The subsequent sections of the study are as follows: Section 2 presents a review of related literature; Section 3 provides an overview of the Malaysian economy and its energy sector dynamics. Section 4 focuses on data and econometrics procedures, Section 5 details the interpretation and discussion of the results, while Section 6 concludes the studies with policy direction for stakeholders, energy regulators, and government officials.
2. Review of related literature
Over the past decades, energy consumption remains the backbone of socio-economic development across economies. This is validated by the seminal work of Kraft and Kraft [12] for the US, which serves as a gateway to several studies in the energy literature for both, single country, cross countries and panel of countries with diverse and insightful outcomes (inter alia [20–22]; Bekun & Agboola,2019; Alola &
Alola,2018 [23–29]; Solarin & Shahbaz,2014; Shahbaz & Lean,2012;
[30,31]. To date, the energy literature has well-documented studies on the trajectory of the energy revolution. However, these studies have focused on the linkage between NG consumption and its effect on eco- nomic output. The literature on NG-economic nexus can be classified into four hypotheses namely (a) growth hypothesis (b) conservative hypothesis (c) feedback hypothesis and (d) neutrality hypothesis. First, the growth hypothesis posits that economic growth drives NG con- sumption — a one-way causality running from economic growth to NG consumption [32]. The second tier reflects on the unidirectional cau- sality from NG to economic growth, known in the energy literature as NG-induced hypothesis (see [33]: — implying that the consumption of NG is a key determinant of economic growth. Meaning that any attempt to apply the conservative hypothesis will hurt such an economy. Third, the feedback hypothesis entails two-way causality running from both NG and economic growth and vice versa (Shahbaz, 2014). Finally, the neutrality hypothesis occurs when there is no causality in either direc- tion from NG consumption and economic growth and vice versa. This means that both variables do not affect each other [11,34]. The appli- cation of the conservative hypothesis can be applied in this situation without an adverse effect on the economy.
There are numerous studies on natural gas — economic growth nexus. For instance, Solarin and Ozturk [28] explored the linkage that exists between NG consumption economic growth for 12 members of the organization of petroleum exporting countries (OPEC) over the period 1980–2012. The study findings for the bloc supports the feedback cau- sality hypothesis. On the contrary, the study for individual countries reported diverse outcomes. For instance, in Nigeria, Kuwait, Iraq, and Saudi Arabia the growth hypothesis was valid while Iran, the United Arab Emirates (UAE), Algeria, and Venezuela join the strands of studies that support the conservative hypothesis. The neutrality hypothesis was confirmed in Angola and Qatar while Ecuador was the only country to supported the bidirectional causality hypothesis (feedback hypothesis).
Zamani [35] investigated the natural-gas induced growth relation- ship in Iran using disaggregated energy consumption through the vector error correction model (VECM) methodology. The study found a feed- back causality between Natural gas consumption and economic growth between 1967 and 2003. Other studies validated the significant role of natural gas on economic growth in Russia, Iran, Qatar, Turkmenistan, and Iran, respectively [36,37]. However, there exists a paucity of studies on the theme for Malaysia — for instance, a study by Solarin and Shahbaz [18]. Our study is in line with Rafindadi and Ozturk [19]; who investigated natural gas-economic growth nexus in a multivariate framework with the inclusion of foreign direct investment (FDI), trade openness and gross capital formation while accounting for a possible structural break. The empirical findings of the study support the feed- back hypothesis between the consumption of natural gas and economic development, FDI and economic development, and natural gas con- sumption and FDI. The line of studies on the theme for selected regions, variables, and hypothesis are reported in Table 1.
3. An overview of the Malaysian economy and its energy dynamics: a brief discourse
Malaysia has a unique geographical feature with a landmass of 329,847 km2 located in the southern Asia Peninsula. With the current population of 32, 386, 784 as of May 2019. This population is equivalent to 0.42% of the total world population. Malaysia’s population density is estimated at 99 per km2 (256 people per mi2). Malaysia operates a constitutional monarchy system that holds thirteen states and three federal territories. The country is bordered around countries like Thailand, in the northern by Indonesia, Brunei and in the South China Sea, south of Vietnam. The Malaysian economy is blessed with natural endowment not limited to petroleum, natural gas, bauxite, iron, copper, ore, timber, etc. Malaysia has gradually transformed its economy from agriculture and commodity, being the producer of raw materials to a Fig. 1. Malaysian energy mix. Data source: US energy information
administration.
global player in the manufacturing and services sector, specifically in the exportation of palm oil, electrical appliances, electronics components, and natural gas as outlined by British Petroleum, (2019).
With the recognition of energy sector as the life wire of the Malaysian economy, deliberate measures in terms of energy policies birthed Acts like the National Depletion Policy of 1980 National Energy Policy of 1979, Petroleum Development Act of 1974, National Petroleum Policy of 1975 and more recently the Energy Commission Act 2001 were adopted to explore and develop a framework for producing petroleum resources, as well as look into issues related to ensuring the continuous supply, utilization and environmental concerns of energy without losing focus to ensure and promote private sector involvement in infra- structural facilities development — all in the bid to stimulate the economy and prolong the existence of the country’s oil reserves [18].
The energy sector given its central role in the Malaysian economy accounts for more than 20% of its GDP. The upstream activities of the oil and gas sector can be estimated as well above RM87 billion whereas the similar activities from the downstream including refining can be esti- mated at more than RM 24 billion. This sector single-handedly accounts for the biggest source of revenue to the Malaysian government through dividends and taxes [60].
The principal forms of the energy consumed in Malaysia include natural gas responsible for 36% and oil account for 40% of the total energy mix. Coal also accounts for about 17% of the total energy mix.
Fig. 1 gives more insights into the energy mix of Malaysia.1
4. Methodology
This section focuses on the econometrics procedures applied, data description, unit of measurement and data source.
4.1. Data
To explore the relationship between NG consumption, gross capital formation, globalization, and CO2 emissions on economic growth in Malaysia, the study constructed a multivariate framework using five variables. The variables include the real gross domestic product (RGDP) used as a proxy for economic growth, gross capital formation used as a proxy for physical capital, carbon dioxide emission, and globalization index as developed by Dreher (2006) which accounts for economic, social and political dimensions of globalization. The intuition behind the choice of variables can be traced from the United Nations Sustainable Development Goals [(UNSDG) 7, 8, 9, 13, and 17] [61].
Natural gas (NG): intentional efforts made in using renewable energy to provide access to electricity and clean cooking fuels constitute a component of the sustainable goals that will enhance growth and sustain the environment (SDG 7).
Economic Growth (RGDP): A high level of productivity is needed to Table 1
Summary of selected studies on natural gas economic growth nexus across the globe.
Author & Year Location Coverage Technique Findings Decision
Khan and Ahmad [33] Pakistan 1972–2007 Johansen test Y → NG Growth
Apergis and Payne [11] 67 Countries 1992–2005 Pedroni cointegration NG ↔ Y Feedback
Yang [38] Taiwan 1954–1997 ARDL, GC, JJ NG ↔ Y Feedback
Adeniran [39] Nigeria 1980–2006 Sims Causality test Y → NG Growth
Zahid [40] India, Pakistan, Bangladesh 1971–2009 TY NG → Y; Y x NG Growth; Neutrality
Farhani et al. [34] Tunisia 1980–2010 ARDL, GC, JJ NG ↔ Y Feedback
Rafindadi and Ozturk [19] Malaysia 1971–2012 ARDL,BH,GC NG ↔ Y Feedback
Shahbaz et al. [10] Pakistan 1972–2010 ARDL,JML,GC NG → Y Conservative
Ighodaro [41] Nigeria 1970–2005 VECM, JJ NG → Y Conservative
Zamani [35] Iran 1967–2003 JML, VECM NG ↔ Y Feedback
Fatai et al. [42] New Zealand and Australia 1960–1999 ARDL, JML, TY Y x NG Neutrality
Solarin and Shahbaz (2014) Malaysia 1971–2012 BH, ARDL, VECM NG ↔ Y Feedback
Hossein et al. [43] OPEC countries 1980–2009 GC Y x NG Neutrality
Kum et al. [44] G-7 Countries 1970–2008 Bootstrap, TY NG → Y; Y x NG Growth; Neutrality
Payne [45] USA 1949–2006 TY Y → NG Growth
Esen and oral (2016) Iran, Russia, Qatar, Turkmenistan N/A Descriptive statistics NG ↔ Y Feedback
Bildirici and Bakirtas [46] Brazil, Russia and Turkey 1980–2011 ARDL, JML,GC NG ↔ Y Feedback
Saboori and Sulaiman [47] Malaysia 1980–2013 ARDL, JML, GC NG ↔ Y Feedback
Aqeel and Butt [48] 1955–1996 Pakistan GC Y x NG Neutrality
Hu and Liu [32] Taiwan 1973–2003 VECM NG ← Y Growth
Akadiri and Akadiri [17] Iran 1980–2013 ARDL, TY Y x NG Neutrality
Furuoka [49] China 1980–2012 ARDL,GC,TY NG → Y Conservative
Das et al. [50] Bangladesh 1980–2010 JML,GC Y → NG Growth
Solarin and Ozturk [28] OPEC member countries 1980–2012 Panel GC NG ↔ Y Feedback
Shahiduzzaman and Alam [51] Australia 1970–2009 ARDL NG ↔ Y Feedback
Ozturk and Al-Mulali [52] Gulf Cooperation Council Countries 1980–2012 Pedroni cointegration test NG ↔ Y Feedback
Pirlogea and Cicea [53] Romania 1990–2010 GC Y x NG Neutrality
Balitskiy et al. [54] EU-26 1997–2011 Panel cointegration NG↔ Y Feedback
Dogan [55] Turkey 1995–2012 VECM, GC NG ↔ Y Feedback
Das et al. [50] Bangladesh 1980–2010 JML,GC Y → NG Growth
Solarin and Lean [56] India and China 1965–2013 Hatemi-J, TYDL GC NG ↔ Y Feedback
Muhammad et al. [57] Pakistan 1972–2010 ARDL NG → Y Conservative
Destek [58] OECD countries 1991–2013 Panel VECM, FMOLS, DOLS NG ↔ Y Feedback
Hafeznia et al. [37] Iran N/A Descriptive stats, Graphs NG ↔ Y Feedback
[59] Iran 1990Q1 - 2017Q4 ARDL, GC, BH NG ↔ Y Feedback
Notes: The definition of the following abbreviations and notations: ↔ feedback causality; → conservative causality; ← growth causality; x no causality, N/A: Not available; NG: Natural Gas; Y: Economic growth; ARDL: Autoregressive Distributed Lag; VECM: Vector Error Correction Model; GC: Granger Causality; BH: Bayer and Hanck; JML: Johansen’s Maximum Likelihood; JJ: Johansen-Joselius Cointegration; TY: Toda and Yamamoto; TYDL: Toda and Yamamoto and Dolado and Luktkepohl.
1 For more insight into energy mix in Malaysia, interested reader may visit the following linkshttps://www.st.gov.my/contents/files/download /116/Malaysia_Energy_Statistics_Handbook_2017.pdfhttps://www.st.gov.my/
achieve full employment in the economy. Hence, SDG 8 seeks to empower entrepreneurs who drive the process, create decent jobs for the massive unemployed population who are ready and able to work. These will help in achieving sustained economic development.
Gross Capital Formation (GCF): The investment needed to build infrastructural facilities will depend on capital formation vis-�a-vis manufacturing and labor productivity. The sum of these will increase investment which will be useful in developing infrastructure, which will, in turn, boost the industrial share of economic development. Hence, helps to promote inclusive, sustainable industrialization and foster innovation (SDG 9).
Carbon dioxide emissions (CO2): The negative effect of greenhouse gas (GHG) emissions on human lives and environment calls for urgent attention, especially when CO2 constitutes a significant portion of GHG.
Therefore, SDG 13 is concerned with reducing climate change hazards and its impact.
Globalization index (GI): The benefits that come from interconnec- tedness and global partnership through massive cooperation and ex- change of ideas are needed to foster economic growth and development.
To attain these, SDG 17 plays a critical role to ensure and enhance access to knowledge and technology with the sole target of achieving this goal.
These variables were sourced from World Bank Development In- dicators (WDI) database and measured in constant 2010 USD for RGDP, GCF, and CO2 in kt whereas NG consumption was derived from the U.S Energy Information Administration database (EIA). The annual data used for the econometric analysis spans from 1980-2014.2 Table 2 de- scribes the data, unit of measurement, and their respective sources.
4.2. Test processes
The study used the following empirical sequence: (a) tested for sta- tionarity among the variables of interest via Augmented Dickey-Fuller (ADF, 1981), Philips Perron (PP, 1988), and Zivot-Andrews [62]. (b) Examined the long-run equilibrium relationship between variables using a combined cointegration test by Bayer & Hanck [63]. The Autore- gressive Distributed Lag Model (ARDL) method of Pesaran et al. [64]
was further explored to test for the robustness of the long-run relation- ship. (c) Granger causality test was carried out to ascertain the direction of causality among variables of interest.
4.3. Model specification
This study is predicated on the existing study of Solarin and Ozturk [65]. Hence, the functional form adopted in the study is expressed as:
RGDP ¼f (NG, GCF, GI, CO2) (1)
In determining that there is homoscedasticity in the variables, log- arithm transformation (ln) was applied to equation (1).
lnRGDPt ¼δ þα1lnNGt þα2lnGCFt þα3lnGIt þα4lnCO2t þϵt (2) Where δ represents intercept or constant and α1, α2, α3, α4 are partial slope parameters to be estimated while ϵt is the stochastic terms to capture unobserved in the fitted model.
4.4. Test of stationarity
The test of stationarity in time series econometrics literature is essential to ascertain the order of integration of a variable before pro- ceeding to test for cointegration and causality test — to prevent spurious analysis and erroneous policy implications. The basic test of stationarity using Elliot et al. [66]; Philips and Perron (1988), and Augment Dickey-Fuller [67] have become inadequate in some sense by lacking the capability to capture structural breaks which are present in most time-series data. This weakens the power of these traditional unit root test to reject the null hypothesis of unit root stationary test. Conse- quently, we introduced a unit root test with structural breaks to com- plement the deficiencies of the traditional unit root test and provide reliable and consistent estimates. Zivot-Andrews [62] was used for this purpose and it is computed as follows:
ΔYt¼β1þβ2tþδYt 1þγDUtþXr
i¼0
ΦiΔYt iþεt (3)
ΔYt¼β1þβ2tþλYt 1þφDTtþ Xr
i¼0
ΦiΔYt iþεt (4)
ΔYt¼β1þβ2tþλYt 1þγDUtþφDTtþXr
i¼0
ΦiΔYt iþεt (5) From equations (3)–(5), Yt refers to the time series examined, Yt-1
denotes the first lag of the time series under consideration and ΔYt-i is lagged first differences to accommodate the serial correlation in the errors. The “t sig” represents the lag length which is useful in producing the test statistics given some information-based criteria. The DUt denotes the dummy variable for the mean shift occurring at each possible structural break date, whereas DTt refers to the dummy variable indi- cator for the trend shift occurring at each possible break date. Equation (3) allows for a unit root test which permits a one time change in the series level. Equation (4) permits the unit root test that allows a one-time change in the slope of the trend function, and finally equation (5) which allows for unit root test by combining one-time changes in the level as well as the slope of the trend function of the series.
The null hypothesis of Zivot-Andrews unit root applies to equations (3)–(5) and it is denoted as θ¼0:Hence, the null hypothesis H0:θ >0 is tested against the alternative of stationarity H1:θ< 0. The null hy- pothesis implies that the series (Yt) contains a unit root with a drift that excludes any structural break, whereas the alternative hypothesis im- plies that the series is a trend-stationary process with a one-time break occurring at a point in time that is unknown. Therefore, in a case where we fail to reject H0 then there is the presence of unit root whereas rejection validates stationarity.
4.5. Measurement of cointegration relationships
The pioneering approach to testing the equilibrium association be- tween variables was advanced by Engle and Granger [68]. The test of cointegration simply requires that variables possess a unique order of integration. Econometric literature reveals that there is a lower inte- gration order especially when the time series are integrated at I(0) or I (1). However, the Engle-Granger cointegration test is puzzled with the challenge of a biased empirical outcome as a result of low explanatory power properties. The Johansen [69] cointegration test is a better option Table 2
Data description and unit of measurement.
Series Unit of Measurement Source
Carbon dioxide (CO2) emissions kt WDI
Real Gross domestic product
(RGDP) Constant 2010 $ USD WDI
Gross capital formation (GCF) Constant 2010 $ USD WDI Globalization index (GI) KOF Index of globalization KOF index Natural gas (NG) Measured in dry NG in billion
ft3 EIA
Author’s compilation.
2 This study coverage span is restricted based on data availability. Also the data for CO2 is available at WDI till 2014. Thus, for balance data and easy of estimations the study is trim to 2014.
of a cointegration test relative to the aforementioned as it allows more than one cointegrating relationship between the variables. The recently advanced Bayer and Hanck [63] cointegration test blend various test statistics ranging from Engle-Granger [68]; Johansen [69]; Boswijk [70]
and Banerjee et al. [71] in the bid to have a robust result from a single framework and arrive at a more robust and comprehensive conclusion.
The current study used the combined cointegration of Bayer-Hanck cointegration test to assess possible cointegration between NG con- sumption and economic growth in Malaysia. Combining the estimated significance level of the individual cointegration test in line with Fishers’ formula is given as:
EG JOH¼ 2flogðP:EGÞ þ ðP:JOHÞg (6)
EG JOH BO BDM¼ 2flogððP:EGÞ þ ðP:JOHÞ þ ðP:BOÞ þ ðP:BDMÞÞg (7) Meaning that,P:EG;P:JOH;P:BOandP:BDM are the individual probabil- ities of each test statistic. That is, p-values of cointegration tests such as Engle-Granger [68]; Johansen (JOH, 1988); Boswijik (BO, 1994) and, Banerjee et al. (BDM, 1998) as represented by P:EG;P:JOH;P:BOandP:BDM
respectively. The decision rule holds that where the calculated Fisher statistics is greater than the critical values provided by the Bayer and Hanck [63]; the null hypothesis of no cointegration can be rejected.
4.6. Autoregressive Distributed Lag (ARDL) approach
The ARDL bounds test approach can be used to revalidate the robustness of the cointegration relationship between NG consumption and economic growth and other variables such as gross capital forma- tion, globalization, and CO2 emissions. This approach guarantees effi- cient estimates especially when the sample size is relatively small compared to other traditional cointegration tests. The ability of this method to report simultaneously the long and short-run dynamics of fitted regression together with error correction model term (ECT) is laudable. Besides the outlined merits, it is also known for its usefulness in the case of an unknown order of stationarity — be it either ~ I(0) or I (1) but certainly not I(2). It is usually estimated within the framework of unrestricted error correction where all the variables are assumed to be endogenous. This estimate is carried as follows:
ΔYt¼δ0þδ1tþβ1yt 1þ XZ
k¼1
γ1vkt 1þ XX
n¼1
ϕnΔYt nþ XZ
k¼1
XX
n¼1
μknΔVkt n
þθDtþεt
(8) The exogenous variable that accommodates the structural breaks in the framework is denoted as Dt whereas the vector is represented by Vk. Where there is no cointegration, the F-statistics computed from the bounds test is used to confirm the null hypothesis. The decision can be made from the following scenarios: (a) a situation where the F-statistics computed is greater than the upper limit of the critical values reported, the null hypothesis of no cointegration is rejected. (b) a case where the F values are within lower and upper bounds, the decision will be incon- clusive, and (c) a situation where the F-statistics is found below the upper limits, the decision, in this case, will be no cointegration.
The bounds test specification is expressed as:
H0: β1 ¼β2 ¼… … ¼βkþ2 ¼0 H1: β1 6¼β2 6¼… …6¼βkþ2 6¼0
Different forms of long-run relationship tests exist in the existing literature that can be employed after validating the presence of cointe- gration among variables. For instance, the among the long run estima- tors includes the Dynamic ordinary least squares, (DOLS), advanced by Ref. [72]; Fully Modified Ordinary Least Squares (FMOLS) proposed by Philips and Hansen (1990) and Park [73] proposed Canonical
Cointegration Regression (CCR).
4.7. Granger Causality approach
Since the traditional regression does not imply causal interaction or association among the variables, it was needful to assess the direction of causality given the marginal benefits for policy formulation. This study used the Granger causality technique to detect the predictability power that exists among the variables of interest.
5. Results interpretations and discussions
The visual plot of the variables depicted in Fig. 2 is essential to un- derstand the trend/patterns of the dataset used in the estimation anal- ysis. Fig. 2 reveals the trend of each series, natural gas, and gross capital formation series exhibit obvious structural breaks relative to gross do- mestic product, globalization index, and CO2 emissions. This study has a provision to capture these structural breaks in subsequent estimation.
Table 3 reports basic descriptive statistics such as averages, variance, minimum, maximum, normality, skewness, and kurtosis. Table 3 shows that economic growth has the highest average followed by real gross fixed capital formation with NG with the lowest mean. Coincidentally over the sampled period, RGDP has maximum with NG minimum. All the series shows significant deviation from their means as revealed by the standard deviation.
Table 4 reports the ADF and PP unit root tests to validate the unit root properties of the series in the study. The results reveal that the series are non-stationary at levels in the presence of structural breaks. How- ever, all the variables are stationary at first difference. Implying that the variables are integrated of I(1) at a 1% level of significance. Further confirmation of ADF is validated by the PP unit root test implying that the order of the integration of the variable is I(1). The fundamental issue with the traditional ADF and PP unit root tests is the inability to capture structural breaks in series thereby resulting in ambiguous and misleading results. However, Zivot and Andrews [62] unit root test can accommodate single unknown structural breaks in the series as observed in Table 5. The Zivot and Andrews unit root test is usually selected in favor of the null hypothesis when considering the break date selection using the t-statistics and it uses the critical values of the ADF unit root test. The identified break dates tally with landmark political and eco- nomic episodes in Malaysian history.
The maximum lag length selection criteria are presented in Table 6, which afford the best model to be selected. The results from Table 6 reveal that the most appropriate criteria for lag length selection are AIC with a lag length of 1 which is capable to accommodate small sample size which is suitable for this nature of the study.
The ARDL short and long-run results to validate the long-run equi- librium relationship are presented in Table 7. The results reveal a slightly above average speed of adjustment of approximately 59% in collaboration with the explanatory variables. In the short run, the empirical results show that a 1% increase in NG consumption leads to an increase in economic output by 0.02% holding other things constant.
This implies that energy (NG) adds to the growth of the Malaysian economy in contrast to a study by Rafindadi and Ozturk [19]. The study reveals a positive and statistically significant relationship with economic output stemming from the increase in energy (NG) consumption.
Meaning that an expansion in NG consumption in Malaysia directly stimulates economic growth. This trend is also found positive and sta- tistically significant in the long run. A positive relationship is observed between gross capital formation and RGDP. An increase in gross capital formation by 1% will increase RGDP by 0.09% — leading to economic expansion. Globalization and CO2 emissions follow a similar positive trend with RGDP. The results in Table 7 further affirm convergence between NG consumption, gross capital formation, globalization, CO2,
and RGDP. The long-run for all the variables is aligned with the trend of the short run. The positive relationship observed between CO2 and
economic growth both in the short and long-run has useful policy im- plications. This increment allows for a tradeoff between environmental quality and economic development. Hence, modernized and environ- mentally friendly energy sources are encouraged especially in the face of a global shift to cleaner energy sources advanced by other nations [74];
Emir & Bekun, 2018). The non-cointegration test results via Bayer and Hanck are reported in Table 8. The null hypothesis of no cointegration is rejected at a 1% significance level, thus, the results confirm a cointe- gration between the variable under consideration. A further step of checking the robustness was considered using the ARDL bounds testing which validated cointegration among variables showed in Table 8.
The FMOLS and CCR estimators were used to investigate the long- run equilibrium elasticities and determine the magnitude of the coin- tegration. This is usually carried out after confirming the existence of cointegration between the variables. Table 9 reveals a positive rela- tionship between NG, GCF, GI, CO2 and RGDP for the two estimators.
This implies that NG, GCF, GI and CO2 are positively related to the dependent variable (RGDP). Both estimation techniques reveal a posi- tive and significant relationship between NG consumption and economic growth. Hence, our estimates validate the growth-induced NG con- sumption hypothesis, as a positive relationship running from RGDP to NG consumption in Malaysia. Besides, a 1% increase in NG consumption will result in a corresponding increase in economic output by 0.028%
and 0.027% for FMOLS and CCR respectively. In the same vein, a pos- itive and statistically significant trend association is observed for gross capital formation, globalization, and CO2 emissions. The positive rela- tionship between globalization, CO2 emissions, and economic growth suggest simultaneous and proper management of the economy and environment. The positive empirical evidence between economic growth and CO2 emissions implies the need for more action to disen- tangle economic growth from environmental pollution, especially from fossil fuel sources. A similar study by Leah and Smith (2009) observed unidirectional causality from energy use to carbon emissions for Iran.
This study further reveals a positive collaboration between gross capital formation and economic growth. This is a clarion call for the Malaysian economy to harness and strengthen institutions on the path of accumulation capital to grow the economy both in the short and long run. This capital accumulation can guarantee sustained economic growth. The results of fitted model residual diagnostic tests reported in Fig. 2.Trend plot of the relationship between Carbon dioxide, economic output, gross capital formation, globalization and natural gas (1981–2014).
Table 3
Descriptive statistics.
lnCO2 lnRGDP lnGCF lnGI lnNG
Mean 11.4673 25.5641 24.2202 4.2131 2.4924
Median 11.7046 25.6754 24.4431 4.2535 2.7601
Maximum 12.4001 26.4737 25.0963 4.3917 3.5493 Minimum 10.2399 24.5469 23.1660 3.9844 0.3947
Std. Dev. 0.7069 0.6008 0.5992 0.1462 0.9381
Skewness 0.3836 0.2090 0.4468 0.3039 0.8335
Kurtosis 1.7143 1.7109 1.8794 1.5198 2.7706
Author’s compilation.
Table 4 Unit root result.
Variables ADF PP
Panel A: Level
lnCO2 1.3187 1.3755
lnRGDP 1.0392 0.9949
lnGCF 1.1139 1.1448
lnGI 1.3294 1.1624
lnNG 2.2940 2.5450
Panel B: Difference
lnCO2 6.3518* 6.3227*
lnRGDP 4.7161* 4.7246*
lnGCF 4.9737* 4.9522*
lnGI 4.2584* 4.2586*
lnNG 4.1843* 4.1740*
Notes:*denotes a rejection of the null at 1% significance level.
Table 10 validate the adequacy of the model for policy direction and guidance. The fitted model is void of violation of any assumption of the classical linear regression model (CLRM) namely serial correlation, heteroscedasticity, and misspecification bias.
Fig. 3 reports the cumulative sum (CUSUM) and the cumulative sum of squares (CUSUMsq) stability diagnostic test of the fitted model.
CUSUM and CUSUMsq tests are essential for examining the stability of short and long-run parameters of NG consumption function. The plots depicted in Fig. 3 are observed within the 95% threshold limit, signi- fying statistical significance at 5%. The implication is that the NG Table 5
Unit root test for single structural break.
Statistics (Level) Statistics (Difference) Conclusion
ZAI ZAT ZAB ZAI ZAT ZAB
lnCO2 3.7763 3.3067 3.8550 7.8608* 6.9301* 8.1548* I (1)
Time Break 1991 1996 1991 1997 1992 1997
Lag Length 1 1 1 1 1 1
lnRGDP 3.1738 2.8636 3.0479 6.2209* 5.0245* 6.2069* I (1)
Time Break 1991 1996 1991 1998 1991 1998
Lag Length 1 1 1 1 1 1
lnGCF 3.0868 2.3875 3.2702 5.9633* 4.9199* 5.8819* I (1)
Time Break 1990 1994 1998 1998 1991 1998
Lag Length 1 1 1 1 1 1
lnGI 3.3536 3.0439 3.2544 6.7284* 6.0440* 6.6015* I (1)
Time Break 1992 2003 1992 1988 1993 1988
Lag Length 1 1 1 1 1 1
lnNG 3.2556 2.1477 3.5229 6.4074* 5.7804* 6.8542* I (1)
Time Break 2007 2008 1989 1987 1990 1991 1990
Lag Length 1 1 1 1 1 1 1
Notes: * represents a 1% significance level. Variables used are in their natural logarithms. ZAB denotes the model with a break in both the trend and intercept; Whereas ZAT and ZAI are for models with a break in trend and intercepts respectively.
Table 6
Lag selection criteria.
Lag LogL LR FPE AIC SC HQ
0 110.8500 NA 1.13e-09 6.4151 6.1884 6.3388
1 285.8914 286.4314* 1.29e-13* 15.5085* 14.1481* 15.0508*
2 305.8952 26.6717 1.96e-13 15.2057 12.7116 14.3665
Notes: (*) denotes lag order selected by the criterion. HQ stands for Hannan Quinn, AIC represents Akaike information criterion, SC denotes Schwarz information criteria, FPE means Final prediction error and lastly LR signifying sequential modified LR statistic.
Table 7
ARDL Long and Short-run result.
lnRGDP ¼f (lnNG, lnGCF, lnGI, lnCO2)
Variable Coefficient Std error t-Statistics Probability Short-run result
ECT ( 1) 0.5853* 0.0905 6.0348 0.0000
ΔlnNG 0.0164*** 0.0087 1.8796 0.0719
ΔlnGCF 0.0945* 0.0315 3.0027 0.0060
ΔlnGI 0.4910* 0.2010 2.4424 0.0220
ΔlnCO2 0.0766 0.0531 1.4422 0.1616
Constant 9.4603* 1.4562 6.4964 0.0000
Long run result
lnNG 0.0280* 0.0111 2.5198 0.0185
lnGCF 0.1615* 0.0378 4.2701 0.0002
lnGI 0.8389* 0.2608 3.2165 0.0036
lnCO2 0.1340 0.1050 1.2472 0.2239
Constant 9.4449* 2.2188 4.2567 0.0003
Notes: Asterisk (*,***) denotes 1%, and 10% significant level, respectively.
Table 8
Bayer and hanck results of non-cointegration.
Fitted Model EG-JOH EG-JOH-BO-
BDM Cointegration
Remark lnRGDP f(lnNGC,
lnRGCF, lnCO2, lnGI) 55.2799*** 56.2988*** Yes ARDL bounds testing to cointegration
Test
Statistic Value Signif. I(0) I(1)
F-statistic 6.2791 10% 3.03 4.06
k 4 5% 3.47 4.57
2.5% 3.89 5.07
1% 4.4 5.72
Notes: The asterisks (***) signifies 1% level of statistical significance. The Critical values of EG-JOH and EG-JOH-BO-BDM are 15.845 and 30.774, respectively.
Table 9
Results of Long run regression (FMOLS and CCR).
Dependent variable: lnRGDP
Variable FMOLS CCR
lnNG 0.0280* 0.0271*
3.5595 3.3272
0.0013 0.0025
lnGCF 0.1900* 0.1837*
11.4843 8.3296
0.0000 0.0000
lnGI 1.1221* 1.0999*
10.2060 8.2938
0.0000 0.0000
lnCO2 0.0061* 0.0226*
3.1334 0.4214
0.0039 0.6767
R2 0.9994 0.9994
Adjusted R2 0.9994 0.9993
S.E. of regression 0.0142 0.0143
Long-run variance 0.0001 0.0001
Mean dependent var 25.5940 25.9540
S.D. dependent var 0.5828 0.5828
Sum squared resid 0.0056 0.0057
Notes: * denotes 1% significance level. Values in bracket denote P-values and []
represents t-statistics.
consumption function reveals the efficiency and consistency of the pa- rameters used in the short and long run as validated by CUSUM and CUSUMsq.
The Granger causality was applied to test the direction of causality among the variables of the model. Such knowledge helps craft appro- priate energy policies for sustainable economic growth. The results presented in Table 11 outline the direction of causality among variables under consideration. The Granger causality helps to detect the predict- ability power of variables by considering the contemporaneous term and its past realization between the underlined variables namely economic growth to NG consumption, gross capital formation, globalization, and CO2 emissions. Table 11 reveals a unidirectional causality stemming from economic growth (RGDP) to NG consumption. This outcome is similar and corroborates the findings of Rafindadi and Ozturk [19]. A similar trend of unidirectional causality is observed running from RGDP to gross capital formation, from globalization to NG consumption and from CO2 to NG consumption. This study gives credence to economic
growth induced NG consumption hypothesis in Malaysia — as causality is observed from RGDP to NG consumption. Also, Globalization and CO2
have no causality on RGDP, whereas a bidirectional causality exists between gross capital formation and NG consumption and CO2 versus gross capital formation. Malaysian economic growth drives NG con- sumption and gross capital formation. This means that energy con- sumption (natural gas) and gross capital formation does not granger cause economic growth in Malaysia. Similarly, the study reveals that globalization and CO2 emissions drive natural gas consumption and not the opposite. This means that globalization and CO2 emissions stimulate an economy that is dependent on natural gas consumption. Thus, deepening diversification of the energy portfolio and capital accumu- lation strategies are essential to enhance sustainable economic growth and ensure a feedback effect on the economy.
Subsequently, we proceeded to explore the impact of one standard deviation shocks on each other through the Impulse response function (IRF). The impulse response function shows the reaction of the depen- dent variable to external impulses from its explanatory variables. It is observed in Fig. 4 that NG consumption is sensitive to economic output, positive and persistent over the entire time horizon. For the response of GDP to gross capital, an inverse and persistent trend are observed over the entire period. The impact of globalization on NG consumption is initially negative from the first 1–3 periods after which turns positive.
This implies that external shocks as a result of changes in the global market have a significant impact on NG consumption in Malaysia. CO2
has a negative impact on NG consumption for the first 2 periods, after which a noticeable persistent impact is observed on NG consumption.
This entails depletion of the environment as such the need to shift to cleaner and friendlier environmental sources like renewables. A feed- back causality is confirmed between NG consumption and real capital formation. We observe that NG consumption is sensitive to positive shocks in real capital formation and vice versa in the long run with persistent impact over the time horizon. Human activities in capital terms contribute to CO2 emissions positively but turn negative and persistent over the time horizon, with a similar pattern in terms of CO2
emissions to gross capital formation.
6. Concluding remarks and policy direction
Due to its lowest carbon intensity, natural gas appears to have the Table 10
Residual diagnostic tests for the fitted model.
Test Coefficient p-Value
Heteroscedasticity (ARCH) 0.1211 0.9412
Normality 2.2100 0.1325
Autocorrelation 0.1441 0.7069
Functional form (Ramsey RESET) 0.2316 0.6347
Author’s compilation.
Fig. 3.CUSUM and CUSUMsq graphical plot.
Table 11
Causality test result.
Null Hypothesis Causality F- Statistic Probability lnNG 6¼>LNRGDP lnRGDP → lnNG 1.4993 0.2390
lnRGDP 6¼>LNNG 6.1350* 0.0028
lnGCF 6¼>lnRGDP lnRGDP → lnGCF 1.1671 0.3420
lnRGDP 6¼>lnGCF 2.8061*** 0.0603
lnGI 6¼>lnRGDP lnGI 6¼lnRGDP 1.9525 0.1470
lnRGDP 6¼>lnGI 1.1241 0.3582
lnCO2 6¼>lnRGDP lnCO2 6¼lnRGDP 1.6271 0.2083
lnRGDP 6¼>lnCO2 1.9834 0.1422
lnGCF 6¼>lnNG lnGCF ↔ lnNG 3.2213** 0.0397
lnNG 6¼>lnGCF 2.5962*** 0.0748
lnGI 6¼>lnNG lnGI → lnNG 8.3079* 0.0005
lnNG 6¼>lnGI 2.0136 0.1377
lnCO2 6¼>lnNG lnCO2 → lnNG 5.5028* 0.0048
lnNG 6¼>lnCO2 1.4096 0.2633
lnGI 6¼>lnGCF lnGI 6¼lnGCF 2.0740 0.1292
lnGCF 6¼>lnGI 0.7652 0.5242
lnCO2 6¼>lnGCF lnCO2 ↔ lnGCF 2.5725*** 0.0767
lnGCF 6¼>lnCO2 2.3289*** 0.0988
lnCO2 6¼>lnGI lnGI 6¼lnNG 2.0413 0.1337
lnGI 6¼>lnCO2 1.9694 0.1444
Notes: Asterisk(s) *,**,*** denote(s) the rejection of the null hypothesis at 1%, 5% and 10% significance levels. The symbol → represents unidirectional cau- sality, ↔ denotes bidirectional causality and 6¼ means neutrality while 6¼>
means does not Granger cause.
Fig. 4.Impulse response graphical plot of shocks among variables.
least environmental effect compared to oil and coal. The discovery has improved Malaysia’s economic development by increasing the total energy consumption. However, studies that examine the natural gas- driven economy is limited, especially in Malaysia. Motivated by the 11th Malaysian Plan and goal 7 of sustainable development target, we investigated the nexus between the consumption of natural gas and economic development for Malaysia from 1980 to 2014 — by inte- grating globalization index, gross capital formation, and CO2 emissions in a multivariate framework. The empirical results showed that a 1%
increase in energy consumption (natural gas) consumption increases economic growth by 0.03% in the long-run and 0.02% in the short-run in Malaysia. This implies that energy consumption (natural gas) con- sumption can help achieve the 11th Malaysian Plan as envisaged. The study further found a short-run gross capital formation-led economic growth by 0.09 and 0.16% in the long-run. Globalization was observed to influence economic development positively by 0.84% with a 1% in- crease in economic growth. Implying that, globalization can signifi- cantly influence economic development in Malaysia. These findings suggest that an increase in the levels of CO2 emissions emanate from human activities such as industrial production, urbanization, trans- portation, and other activities that translate into higher economic output both in the short- and long-run. The dual findings of natural gas consumption and gross capital formation influencing economic output have policy implications. Based on the empirical revelations, the following policy directions are made:
(i) The study confirmed the growth-induced natural gas consump- tion hypothesis as highlighted by the Granger causality analysis.
Such causality insights help craft appropriate energy policies for sustainable economic growth.
(ii) The role of capital in economic growth is significant to the Malaysian economic prosperity, hence, promote inclusive, sus- tainable industrialization and foster innovation (SDG 9) as sug- gested in the causality result. As such, policymakers are enjoined to intensify efforts to increase both physical and human capital accumulation in the country to achieve the 7th Sustainable Development Goal (SDG) and the agenda 2020 of the 11th Malaysian Plan.
(iii) The need for the Malaysian government to strengthen its in- stitutions on environmental treaties and regulations like the Kyoto Protocol and Paris Agreement is a requirement to check its emission level, thereby contributing to the reduction of the global average temperature of below 1.5 �C.
This study serves as a beacon to other Asian countries in their quest to improve economic growth without tradeoff for environmental quality and achieving SDGs 7 and 8 and the 11th Malaysian Plan. Future research needs to revisit the theme by considering other growth drivers like population, democratic regime, and good governance in terms of asymmetry, given the paucity of studies.
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
CRediT authorship contribution statement
Mfonobong Udom Etokakpan: Conceptualization, Data curation, Formal analysis, Methodology, Software, Validation, Visualization, Writing - review & editing. Sakiru Adebola Solarin: Writing - original draft. Vedat Yorucu: Writing - original draft. Festus Victor Bekun:
Writing - original draft. Samuel Asumadu Sarkodie: Writing - original draft, Writing - review & editing, Funding acquisition.
Acknowledgments
Open Access funding provided by Nord University.
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://doi.
org/10.1016/j.esr.2020.100526.
References
[1] EIA, Energy InformationAdministration(EIA), 2018. Available at: https://www.eia.
gov/outlooks/ieo. (Accessed 15 April 2019).
[2] R. Samu, F.V. Bekun, M. Fahrioglu, Electricity consumption and economic growth nexus in Zimbabwe revisited: fresh evidence from Maki cointegration, Int. J. Green Energy 16 (7) (2019) 540–550.
[3] S.A. Sarkodie, S. Adams, Renewable energy, nuclear energy, and environmental pollution: accounting for political institutional quality in South Africa, Sci. Total Environ. 643 (2018) 1590–1601.
[4] S. Asumadu-Sarkodie, P.A. Owusu, Carbon dioxide emissions, GDP, energy use, and population growth: a multivariate and causality analysis for Ghana, 1971–2013, Environ. Sci. Pollut. Control Ser. 23 (13) (2016) 13508–13520.
[5] P.A. Owusu, S.S. Asumadu, A review of renewable energy sources, sustainability issues and climate change mitigation, Cogent Engineering 3 (1) (2016) 1167990.
[6] S.A. Solarin, M. Shahbaz, Trivariate causality between economic growth, urbanisation and electricity consumption in Angola: cointegration and causality analysis, Energy Pol. 60 (2013) 876–884.
[7] M.U. Etokakpan, F.F. Adedoyin, V. Yorucu, F.V. Bekun, Does globalization in Turkey induce increased energy consumption: insights into its environmental pros and cons, Environ. Sci. Pollut. Res. 27 (2020) 26125–26140, https://doi.org/
10.1007/s11356-020-08714-3.
[8] J.W. Tester, E.M. Drake, M.J. Driscoll, M.W. Golay, W.A. Peters, Sustainable Energy: Choosing Among Options, MIT press, 2012.
[9] British Petroleum, Available in Statistical Review of World Energy | Energy Economics, 2019. https://www.bp.com/en/global/corporate/energy-economics/
statistical-review-of-world-energy.html.
[10] M. Shahbaz, H. Lean, A. Farooq, Natural gas consumption and economic growth in Pakistan, Renew. Sustain. Energy Rev. 18 (2013) 87–94.
[11] N. Apergis, J.E. Payne, Natural gas consumption and economic growth: a panel investigation of 67 countries, Appl. Energy 87 (8) (2010) 2759–2763.
[12] J. Kraft, A. Kraft, On the relationship between energy and GNP, J. Energy Dev.
(1978) 401–403.
[13] A.A. Alola, The trilemma of trade, monetary and immigration policies in the United States: accounting for environmental sustainability, Sci. Total Environ. 658 (2019) 260–267.
[14] O. Usman, P.T. Iorember, I.O. Olanipekun, Revisiting the environmental Kuznets curve (EKC) hypothesis in India: the effects of energy consumption and democracy, Environ. Sci. Pollut. Control Ser. (2019) 1–11.
[15] S.A. Solarin, U. Al-Mulali, I. Musah, I. Ozturk, Investigating the pollution haven hypothesis in Ghana: an empirical investigation, Energy 124 (2017) 706–719.
[16] S.A. Solarin, Electricity consumption and economic growth: trivariate investigation in Botswana with capital formation, Int. J. Energy Econ. Pol. 1 (2) (2011) 32–46.
[17] S.S. Akadiri, A.C. Akadiri, The Role of NG Consumption in Economic Growth, Strategic Planning for Energy and the Environment, 2019.
[18] S.A. Solarin, M. Shahbaz, NG consumption and economic growth: the role of foreign direct investment, capital formation and trade openness in Malaysia, Renew. Sustain. Energy Rev. 42 (2015) 835–845.
[19] A.A. Rafindadi, I. Ozturk, Natural gas consumption and economic growth nexus: is the 10th Malaysian plan attainable within the limits of its resource? Renew.
Sustain. Energy Rev. 49 (2015) 1221–1232.
[20] A.A. Alola, K. Yalçiner, U.V. Alola, S. Saint Akadiri, The role of renewable energy, immigration and real income in environmental sustainability target. Evidence from Europe largest states, Sci. Total Environ. (2019) 1–15.
[21] F.V. Bekun, F. Emir, S.A. Sarkodie, Another look at the relationship between energy consumption, carbon dioxide emissions, and economic growth in South Africa, Sci.
Total Environ. 655 (2019) 759–765.
[22] F.V. Bekun, A.A. Alola, S.A. Sarkodie, Toward a sustainable environment: nexus between CO2 emissions, resource rent, renewable and nonrenewable energy in 16- EU countries, Sci. Total Environ. 657 (2019) 1023–1029.
[23] A.C. Akadiri, S. Saint Akadiri, H. Gungor, The role of natural gas consumption in Saudi Arabia’s output and its implication for trade and environmental quality, Energy Pol. 129 (2019) 230–238.
[24] S.S. Akadiri, F.V. Bekun, E. Taheri, A.C. Akadiri, Carbon emissions, energy consumption and economic growth: a causality evidence, Int. J. Energy Technol.
Pol. 15 (2–3) (2019) 320–336.
[25] S.S. Akadiri, M.M. Alkawfi, S. U�gural, A.C. Akadiri, Towards achieving environmental sustainability target in Italy. The role of energy, real income and globalization, Sci. Total Environ. (2019) 1–13.
[26] S.A. Sarkodie, P.K. Adom, Determinants of Energy Consumption in Kenya: a NIPALS, 2018.
[27] S.A. Sarkodie, A.O. Crentsil, P.A. Owusu, Does energy consumption follow asymmetric behavior? An assessment of Ghana’s energy sector dynamics, Sci. Total
